doi: 10.1002/lno.11731
Seasonal shifts of microbial methane oxidation in Arctic shelf waters above gas seeps
Friederike Gründger ,
1,2* David Probandt ,
3Katrin Knittel,
3Vincent Carrier ,
1,4Dimitri Kalenitchenko ,
1Anna Silyakova ,
1Pavel Serov ,
1Bénédicte Ferré ,
1Mette M. Svenning ,
1,4Helge Niemann
5,6,11CAGE—Centre for Arctic Gas Hydrate, Environment and Climate, Department of Geosciences, UiT The Arctic University of Norway, Tromsø, Norway
2Arctic Research Centre, Department of Biology, Aarhus University, Aarhus, Denmark
3Department of Molecular Ecology, Max Planck Institute for Marine Microbiology, Bremen, Germany
4Department of Arctic and Marine Biology, UiT The Arctic University of Norway, Tromsø, Norway
5Department of Marine Microbiology & Biogeochemistry, NIOZ Royal Netherlands Institute for Sea Research, Texel, The Netherlands
6Department of Earth Sciences, Faculty of Geosciences, Utrecht University, Utrecht, The Netherlands
Abstract
The Arctic Ocean subseabed holds vast reservoirs of the potent greenhouse gas methane (CH4), often seeping into the ocean water column. In a continuously warming ocean as a result of climate change an increase of CH4
seepage from the seabed is hypothesized. Today, CH4is largely retained in the water column due to the activity of methane-oxidizing bacteria (MOB) that thrive there. Predicted future oceanographic changes, bottom water warming and increasing CH4release may alter efficacy of this microbially mediated CH4sink. Here we investi- gate the composition and principle controls on abundance and activity of the MOB communities at the shallow continental shelf west of Svalbard, which is subject to strong seasonal changes in oceanographic conditions.
Covering a large area (364 km2), we measured vertical distribution of microbial methane oxidation (MOx) rates, MOB community composition, dissolved CH4concentrations, temperature and salinity four times throughout spring and summer during three consecutive years. Sequencing analyses of thepmoAgene revealed a small, rela- tively uniform community mainly composed of type-Ia methanotrophs (deep-sea 3 clade). We found highest MOx rates (7 nM d−1) in summer in bathymetric depressions filled with stagnant Atlantic Water containing moderate concentrations of dissolved CH4(< 100 nM). MOx rates in these depressions during spring were much lower (< 0.5 nM d−1) due to lower temperatures and mixing of Transformed Atlantic Waterflushing MOB with the Atlantic Water out of the depressions. Our results show that MOB and MOx in CH4-rich bottom waters are highly affected by geomorphology and seasonal conditions.
Temperature rise in the Arctic and its impact on the envi- ronment is more severe than for any other region on Earth (Masson-Delmotte et al. 2006; Hansen et al. 2013). The Arctic Ocean holds vast reservoirs of CH4, which has a 32-fold higher greenhouse warming potential than carbon dioxide and may be released into the ocean and the atmosphere (Etminan et al. 2016). The majority of the CH4reservoirs across the Arc- tic shelves are temperature-sensitive, e.g., subsea permafrost (Shakhova et al. 2010) and gas hydrates in shallow sediments (Westbrook et al. 2009; Berndt et al. 2014). Gaseous CH4 released from the seafloor becomes dissolved and can then be utilized by aerobic methane-oxidizing bacteria (MOB), which use CH4 as an energy source and carbon substrate for growth (e.g., Hanson and Hanson 1996; Murrell 2010). In the ocean, aerobic microbial methane oxidation (MOx) is thefinal sink for
*Correspondence: friederike.gruendger@bio.au.dk
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
Additional Supporting Information may be found in the online version of this article.
Author Contribution Statement:F.G. and H.N. designed the project.
F.G., H.N., V.C. collected the samples, carried out the main experiments.
A.S. provided and analyzed hydrographic data. D.P., D.K., V.C., and K.K. carried out bioinformatic and quantification analyses. P.S. analyzed CH4concentrations. M.S. and H.N. supervised the research and contrib- uted with scientific discussions. F.G. carried out interpretation of the experiments and wrote the manuscript with contribution from all co- authors.
CH4 before it is liberated to the atmosphere (Reeburgh 2007), but little is known about the diversity, abundance, distribution, and activity of MOB (Tavormina et al. 2008; Mau et al. 2013;
Steinle et al. 2015). For example, high amounts of CH4were rapidly consumed by MOB following the deep-water horizon accident (Crespo-Medina et al. 2014) and MOB were found to effectively consume CH4 from the water column if hydro- graphic conditions provide continuity for MOB (Steinle et al. 2015, 2017). However, MOx can also be very low despite high CH4concentrations in marine waters for reasons that are still unclear (Bussmann 2013).
Aerobic MOB are phylogenetically divided into Gammaproteobacteria (type I MOB), Alphaproteobacteria (type II MOB) (e.g., Hanson and Hanson 1996; Knief 2015), and Ver- rucomicrobia (type III MOB) (Dunfield et al. 2007; Op den Camp et al. 2009; van Teeseling et al. 2014). Common to almost all MOB is the presence of the membrane-bound
particulate methane monooxygenase, the enzyme responsible for the initial step of methane oxidation. The highly conserved pmoA gene, encoding a subunit of the particulate methane monooxygenase, is most frequently used as a molecular marker both for detection and phylogeny of MOB via cultivation- independent methods (Tavormina et al. 2008; Knief 2015).
Potential methane-oxidizing uncultivated clades like the deep- sea clades 1–5 (Lüke and Frenzel 2011) have been identified by this approach. Especially in marine environments, great uncer- tainties exist about the factors that determine MOB activity and community structure. From the Arctic marine environ- ment, a number of studies report on MOx activity and MOB community composition in relation to environmental factors such as CH4 concentrations and hydrography (e.g., Mau et al. 2013; Steinle et al. 2015; Osudar et al. 2016). However, all those studies are single snap-shots of the prevailing situation at the place and time of sampling. For example, studies focusing
Fig 1.Bathymetric map of the study areas west off Svalbard archipelago showing hydrographic sampling stations indicated by black dots at Isfjorden (Stas.
I–X), Outer Bellsundet (Sta. XI), Outer Hornsund (Sta. XII), and Sørkappøya (Sta. XIII) (A). Detailed map of the shallow shelf west of Prins Karls Forland including gasflare locations (white dots) and 64 sampling stations arranged in a grid. At stations along the four transects (conducted from North to South and West to East, yellow dashed lines), sampling for methane oxidation rates and microbial molecular analyses were conducted in addition to hydrographic profiles and CH4concentrations, which were usually conducted at all stations (B). Global view of our sampling area at the western Svalbard margin (C).
on areas around the Svalbard archipelago, i.e., the continental slope west of Prins Karls Forland (Gentz et al. 2014), Storfjorden in the south-east (Mau et al. 2013), and the Svalbard margin between Bjørnøya and Kongsfjordrenna (south to west Svalbard) (Mau et al. 2017), were only conducted without temporal repeti- tion in August/September. Only Steinle et al. (2015) compared spatiotemporal replicates collected during several surveys at the Svalbard continental margin in late August. Time series studies covering seasonal changes of CH4input and microbial CH4turn- over in the water column in the Arctic have so far not been publi- shed, and little knowledge exists on the community composition of Arctic Ocean MOB (James et al. 2016; Ferré et al. 2020).
In this study, we investigate the fate of CH4 in the water column at the continental margin west of Svalbard, specifi- cally the shallow shelf west off Prins Karls Forland. This area is characterized by CH4 seepage (Portnov et al. 2016; Silyakova et al. 2020). Despite extensive release of CH4 from the sedi- ment in this area, almost no CH4 was found to reach the atmosphere (Myhre et al. 2016). Our main objectives were (1) to study the composition and activity of the methanotrophic communities, (2) to investigate seasonal shifts and, related to these, (3) the differential hydrographical settings and their influence on MOB activity and distribution within the study area. To meet these aims, we conducted sam- pling surveys in the spring, late spring and summer.
We found that community changes of MOB are marginal, but that MOx capacity is influenced by seasonal shifts and varies according to site-specific geographical features and changing hydrographical conditions.
Methods
Study area
Our study area stretches along the continental margin off western Svalbard from the shallow shelf west of Prins Karls Forland towards the southern tip of Svalbard including Isfjorden, Isfjorden Trough, Outer Bellsundet, Outer Hornsund, and Sørkappøya (Fig. 1). Water depth in these areas ranges from 50 to 160 m. The shallow shelf west of Prins Karls Forland is characterized by an irregular bathymetry showing numerous large depressions encompassed by a series of moraine ridges termed the Forlandet moraine complex (Landvik et al. 2005; Fig. 1B). Here, along the Forlandet moraine complex in 80–90 m water depth, a vast number of gas flares (~200flares, identified by acoustic signatures of gas bubbles in the water) were previously mapped (Sahling et al. 2014; Silyakova et al. 2020). Theδ13C values of the emit- ted CH4 and the absence of higher hydrocarbons in the seeping gas indicates a microbial CH4 origin (Graves et al. 2017; Mau et al. 2017). The seepage region west of Prins Karls Forland lies > 200 m shallower than the upper limit of the methane hydrate stability zone and unlikely results from CH4hydrates dissociating in situ. However, lateral migration of CH4from a hypothesized gas hydrate dissociation front at
deeper shelf settings may at least partly fuel the seeps on the shallow shelf (Sarkar et al. 2012).
The hydrodynamics in our study area are complex (see also Silyakova et al. 2020). The East Spitsbergen Current flows along the Svalbard islands southwards on the east side, follow- ing the coast around the island’s southern tip and then turns northwards on the west side of the island (Nilsen et al. 2008).
Here, itflows as a coastal current on the shelf and is composed of less saline and cold Arctic Water (34.30–34.80,−1.5 to 1.0C) into our study area. To the west of the shelf, the northernmost extension of the North Atlantic Current, the West Spitsbergen Current (Aagaard et al. 1987) is composed of relatively saline and warm Atlantic Water (> 34.65, > 3.0C) and also flows northward. Although the East Spitsbergen Current and West Spitsbergen Current are separated by a front, frequent mixing occurs and the West Spitsbergen Current may also flood the shelf (Steinle et al. 2015). Seasonality defines the different por- tions of mixed water masses. Atlantic Water transforms into Transformed Atlantic Water (> 34.65, 1.0–3.0C) by losing heat to the atmosphere and adjacent waters and freshening due to meltwater from glaciers, snow and sea ice. Whereas Intermediate Water (34.00–34.65, > 1.0C) is formed by entrainment and mixing mechanisms at the boundary of Surface Water with underlying Atlantic Water or Transformed Atlantic Water. Sur- face Waters are freshened by melt water and warmed by solar heat in summer (< 34.00, < 1.0C). Waters that overwinter in fjords become colder and fresher, and are then classified as Local Water mass (34.30–34.85,−0.5 to 1.0C). Water masses are clas- sified according to Cottier et al. (2005).
Sampling strategy
Samples were taken within three successive years (2015–2017) during four expeditions with R/VHelmer Hansen,
Table 1. Sampling strategy and definitions of water samples/
horizon taken from the water column.
Water depth CH4 MOx 16S rRNA
Water level
Water layer 5 m below sea
surface
x x x 8 Surface
15 m below sea surface
x x 7
25 m below sea surface
x x 6
Intermediate 2 x x x 5 Intermediate
Intermediate 1 x x 4
25 m above seafloor
x x x 3 Bottom
15 m above seafloor
x x 2
5 m above seafloor
x x x 1
CAGE16-4 (spring: 2–4 May 2016), CAGE17-1 (spring: 16–20 May 2017), CAGE16-5 (late spring: 17–22 June 2016), and CAGE15-3 (summer: 1–3 July 2015). At the shallow shelf west of Prins Karls Forland, we collected samples at 64 hydro- graphic stations arranged in a grid pattern covering an area of approximately 14×26 km. All stations were sampled suc- cessively (total sampling time of the entire grid < 72 h). The positions of transects and stations were selected according to their coverage of specific features found at the seafloor such as gas flares or bathymetric depressions (Fig. 1B). For example, the western N-S transect follows the main ridge of the Forlandet moraine complex and covers numerous CH4
flares. The southern W-E transect features aflare cluster in the west and covers bathymetric depressions towards the east. During the June survey in 2016, time constrains allowed us to only sample 12 of the 64 stations. Those 12 stations were located along the western N-S, eastern N-S, and southern W-E transects (Fig. 1B, Table S1).
At all stations, hydrographic parameters (salinity, tem- perature, pressure) were recorded at 24 Hz with a Conductivity-temperature-depth profiler (SBE 911 plus CTD; Sea-Bird Electronics, Inc., USA) equipped with twelve 5-liter Teflon-lined Niskin bottles. Only downcasts were used for hydrographic profiling. With the CTD-mounted Niskin bottles, we collected discrete water samples from eight water levels (1–8): 5, 15, 25 m above seafloor, 5, 15, 25 m below sea level, and two additional intermediate sam- pling levels evenly spaced between 25 m below sea level and 25 m above seafloor (actual sampling depth depending on water depth; Table 1). Water from the Niskin bottles was subsampled immediately upon recovery. Dissolved CH4
concentrations were measured in all eight water levels of the entire sampling grid. MOx rates were measured in all eight water levels along four transects, which run from north to south (eastern N-S and western N-S) and west to east (northern W-E and southern W-E) (comprising 31 sta- tions, Fig. 1B). Samples for phylogenetic analyses were recovered during sampling campaigns in July 2015 as well as May and June 2016 (but not in May 2017) from water levels 1, 2, 3, 5, and 8 from all stations where MOx rates were measured. During the May 2016 campaign, six stations (Stas. 9, 16, 31, 44, 54, and 64) along the southern W-E transect were repeatedly investigated 2 days after the first sampling to monitor rapid variations of hydrographic con- ditions and their effects on MOx activity and bacterial com- munity changes.
In addition to the shallow shelf west of Prins Karls Forland, in May and June 2016 we investigated the water column at Isfjorden (Stas. I and II), Isfjorden Trough (III–X), Outer Bellsundet (XI), Outer Hornsund (XII), and Sørkappøya (XIII) (Fig. 1A, Table S1) in the same manner as described above.
CTD and dissolved CH4 concentration data from the surveys in July 2015 and June 2016 of the area west of Prins Karls Forland were published by Silyakova et al. (2020).
Methane concentration measurements
Methane concentrations were determined using a head- space method as described by Silyakova et al. (2020). In order to calculate the content of dissolved CH4 of the entire sam- pling area at the shallow shelf of Prins Karls Forland, we defined three water layers with consideration of the uneven bathymetry (Silyakova et al. 2020). Water layers were defined as the"Bottom Water Layer”from seafloor to 25 m above sea- floor (comprising levels 1, 2, and 3), the“Surface Water Layer” from the ocean surface down to 25 m depth (comprising levels 8, 7, and 6), and the ”Intermediate Layer” between 25 m below the ocean surface and 25 m above seafloor (comprising levels 5 and 4) (Table 1).
Methane oxidation rate measurements
Methane oxidation rates were determined according to pre- vious publications (Niemann et al. 2015; Steinle et al. 2015) with modifications as described in Ferré et al. (2020). For the water column at the sampling area west of Prins Karls Forland, the mean areal turnover of CH4was calculated by integrating distinct MOx rates over depth yielding results in m−2d−1for each water layer and the entire water column (Steinle et al. 2017). We then calculated weighted MOx means for each layer, considering uneven horizontal spacing of the hydro cast stations (for a more detailed description of the computation of the weighted means see Silyakova et al. (2020)). Upscaled to the size of the sampling grid (423 km2), these weighted means translate to a total CH4turn- over per day for each water layer and the entire water body of the sampling grid. To compare the capacity of MOx to retain CH4, we then calculated the fraction of CH4 consumed per day:
CH4turnover per dayð Þ% =MOx=CH4×100 ð1Þ
Bacterial community analyses
Seawater samples for molecular analysis were collected in sterile, high-density polyethylene bottles and usually processed immediately after subsampling. However, time con- straints sometimes required storage of samples at 4C in the dark before further processing, but storage time never exceed 4 h. Wefiltered a volume of 1 liter of sample on membranefil- ters (Whatman Nuclepore Track-Etched PC, 0.22 μm, Merck Millipore, MA) by applying a gentle vacuum of ~ 0.5 bar and storedfilters at−20C until further analyses. Total DNA from membrane filters was extracted following the method of Pilloni et al. (2012) and DNA content in each sample was quantified using a spectrophotometer (Nanodrop, ND-1000, Thermo Scientific, MA).
For the amplification of the bacterial 16S rRNA gene, we selected samples retrieved in July 2015, May 2016 and June 2016 from distinct water levels from Prins Karls Forland (Stas.
9, 10, 19, 49, 54, 58; levels 1, 3, 5, and 8), from Isfjorden (Sta.
I; levels 1 and 3) and Outer Bellsundet (Sta. XI; levels 2 and 4) (Table S1). We used the degenerated primer pair Bakt_341F and Bakt_805R resulting in about 450 bp amplicons covering the V3-4 region of the 16S rRNA gene (Herlemann et al. 2011).
For the amplification of the particulate methane mono- oxygenase gene (pmoA), we selected samples from the above- mentioned sampling campaigns from the water level 1 from Prins Karls Forland stations 9, 10, 49, 54, 58, and Isfjorden (Sta.
I), and from water level 2 from Outer Bellsundet (Sta. XI). Water levels 1 and 2 were selected, because of elevated MOx rates, which were measured at these particular water levels at the cho- sen stations. The primer pair wcpmoA189f and wcpmoA661r for marine water column MOB (Tavormina et al. 2008) was used for amplification. Gene analyses of 16S rRNA andpmoA amplicons were performed by IMGM Laboratories GmbH (Martinsried, Germany). Cluster generation and bidirectional sequencing (2 ×300 nt) by synthesis was accomplished on the Illumina MiSeq next generation sequencing platform (Illumina, CA) using reagents kit 500 cycles v2 under the control of MiSeq Control Software v2.5.1.3. Sequence data have been submitted to the GenBank Sequence Read Archives (https://www.ncbi.nlm.nih.
gov/sra) under BioProject ID: PRJNA642858 (16S rRNA) and PRJNA641979 (pmoA).
Processing of 16S rRNA gene sequences was achieved as out- lined in the following. Using BBMerge software version 37.02 (BBTools package, Brian Bushnell, Walnut Creek, CA), we per- formed the initial data processing of the raw sequences (primer trimming, quality filtering at a minimum of 99.1% base call accuracy, and read assembly of forward and reverse read with an overlap of 20 bp by default strictness setting). The resulting reads were processed according to the MiSeq standard operating proce- dure (Kozich et al. 2013) with MOTHUR version 1.39.5 (Schloss et al. 2009) including pre-clustering at 99% sequence similarity and de novo-based chimera removal using UCHIME (Edgar et al. 2011) to remove artificial diversity. Sequences were aligned and classified using the SILVA reference database (Release 132;
Quast et al. 2013). Sequences classified as"no relative,”chloro- plast, archaeal and eukaryotic 16S rRNA were removed. After- wards, bacterial sequences were clustered into operational taxonomic units (OTUs) at 97% sequence similarity, using the OptiClust algorithm. To reduce artificial diversity, rare OTU0.97
that were represented by only≤2 sequences in the whole dataset were removed as suggested for short fragment 16S rRNA gene data (Allen et al. 2016). Prior to diversity analysis, OTUs retrieved for the blank DNA extraction and the no template neg- ative control from the library preparation were also removed from the entire data set.
After MOTHUR sequence processing and prior to statistical and diversity analysis, the community dataset was randomly rarified to the lowest number of sequences found per sample (1083). We conducted alpha and beta diversity analyses using the program R with the vegan package (v. 2.5-6; https://
CRAN.R-project.org/package=vegan). Displaced alpha diversity values are the means of 25 iterations. Hierarchical clustering
and Non-Metric Multidimensional Scaling (NMDS) analyses are based on the Bray–Curtis dissimilarity index. Spearman’s rank correlation of bacterial phyla at the family level and selected environmental variables was conducted with R pack- age Hmisc (v.4.4-0; https://CRAN.R-project.org/package=
Hmisc). Only those family level clades (SILVA taxonomy v.132) that contributed≥0.5% to total sequences in at least one sample were considered.
We processedpmoAamplicons as follows: The raw sequences were treated following our open-access pipeline (https://github.
com/dimikalen/MS_UIT_CAGE/blob/master/CAGE_MiSeq_SOP.
sh). Briefly, forward and reverse reads were merged using BBmerge (v37.36; Bushnell et al. 2017) and qualityfiltered with a maxEE parameter of 1 in VSEARCH (v2.9.0; Rognes et al. (2016)). To reduce the computational need in the following steps, unique sequences were extracted. Operational phyloge- netic units (OPUs) were defined by using USEARCH (v11;
Edgar (2010)) applying a similarity threshold of 97%. The most abundant reads of each OPU were then selected tofind the clos- est known sequences in the pmoA gene reference database (on nucleotide level, Wen et al. (2016)) using the Wang method in MOTHUR (v1.39; Schloss et al. 2009). All raw pmoA reads were then mapped back to the reference reads of the pmoA OPU0.97, as recommended in the USEARCH documentation, in order to construct thefinal OPU table. The OPU table was subse- quently rarefied at 1600 sequences. Hierarchical clustering based on the weighted Unifrac distance metric (Lozupone et al. 2011) was computed by using Qiime (Caporaso et al. 2010) and visuali- zations were made in R (stats package v3.6.1). For phylogenetic analysis of the two most abundant OPUs, we selected 52pmoA sequences from cultured and uncultured MOB published by Lüke and Frenzel (2011), Knief (2015) and in the NCBI GenBank (https://www.ncbi.nlm.nih.gov/). Sequences were aligned using MUSCLE implemented in MEGA 7 (Edgar 2004) and trimmed to retain only shared base pair positions. We built a best-scoring maximum likelihood phylogenetic tree (based on nucleotides) in Randomized Axelerated Maximum Likelihood (RAxML, version 8.2) using the General Time Reversible (GTR) Gamma model (Stamatakis 2014). Thereafter, OPU1 and 2 were aligned to the previously selected sequences and placed into the built phyloge- netic tree using the Evolutionary Placement Algorithm implemented in RAxML. The resulting tree was visualized and annotated in Interactive Tree Of Life (Letunic and Bork 2016).
Results
Our data were collected during sampling surveys in the Arc- tic spring (May 2016, 2017, the month with the coldest bot- tom water temperatures; Berndt et al. (2014)), late spring (June 2016) and summer (July 2015). We mainly investigated a large seep area at the shallow shelf west of Prins Karls Forland and six additional regions, which are hydrographically connected to the Prins Karls Forland shelf area: Isfjorden (I–VII), Isfjorden Trough (VIII–X), and three stations towards the southern tip
of Svalbard: Outer Bellsundet (XI), Outer Hornsund (XII), and Sørkappøya (XIII). We collected discrete water samples for geo- chemical and microbiological analyses from eight defined water levels, whereas CTD measurements were made continu- ously throughout the water column. To simplify the following discussion about comparing processes at the ocean surface or close to the seafloor, we use the three main water layers,"Bot- tom Water Layer,” ”Surface Water Layer” and ”Intermediate Layer”(as defined previously) in order to account for the dif- ferential water depth at the various stations.
Hydrographic setting
In May 2016, the entire water column was dominated by Transformed Atlantic Water with high salinity of 34.9–35.0 and temperatures between 1.6 and 2.3C (Fig. 2A–D). In
contrast, the water column in May 2017 was dominated by relatively warm (2–4.8C) and saline (34.6–34.8) Atlantic Water. Water in the bathymetric depressions was slightly col- der (2.3–2.8C) and therefore classified as Transformed Atlan- tic Water (Fig. 2E–H). In June 2016, the bottom water was composed of Transformed Atlantic Water (lowest temperature 2.3C) at the gasflare area (southern part of the western N-S transect) and within the bathymetric depressions, while we found warmer Atlantic Water (3–5C) in the upper water col- umn. Both water masses were characterized by salinities of 34.6–34.9 (Fig. 2I–K). A strong stratification was observed in July 2015 (Fig. 2L–O). At the bottom of the water column, spe- cifically in bathymetric depressions, water was saline and rela- tively warm Atlantic Water (34.9, ~3.5C) with fractions of Transformed Atlantic Water admixture. At the main gasflare
Fig 2.Profiles of potential temperature in the water column along transects at the shallow shelf west of Prins Karls Forland (A–O) from four sampling sur- veys in May, June, and July within three successive years (2015–2017). For each sampling survey, water depth on they-axis is given in meters below sea level (mbsl). Vertical lines represent stations for continuous CTD measurements. The color code shows measured and linearly interpolated temperature values (C). Selected salinity horizons (values in psu) are indicated by black lines inL–O. In May and June (A–K), no salinity horizons were observed due to a well-mixed water column with constant high salinity levels (34.6–35.0). Each plot contains the bathymetrical baseline (black line above gray area) characteristic for each transect. In June 2016, only three transects were conducted.
area, cold Intermediate Water (34–34.3, 1.3–2C) lay above warmer Atlantic Water. Surface Water with temperatures of 4.5–5.5C and low salinity (28.5–34.0) dominated the surface water down to ~20 m (Fig. 2L–M). The detailed hydrographic setting in May 2016 and July 2015 at the shelf west of Prins Karls Forland has been described by Silyakova et al. (2020).
At Isfjorden and Isfjorden Trough in May 2016, surface water temperatures were colder compared to the Prins Karls Forland shelf, i.e., between−0.2 and 1.2C and with average salinities around 34.6, indicating Local Water and Intermedi- ate Water formed due to local cooling and freshening over winter. Below the surface water layer, water salinity and tem- perature increased (> 34.86, 2.2–2.7C) indicating that the Atlantic Water transformed into Transformed Atlantic Water in the fjord (Fig. S1A). The water column at the central sta- tions of the Isfjorden Trough crossing transect (Stas. IV and V), was characterized by relatively warm Atlantic Water (35,
3.3C) at the surface, which decreased in salinity and tempera- ture to 34.8 and 1.6C at 300 m water depth turning into Transformed Atlantic Water, indicating that at deeper depths, water masses were strongly influenced by mixing with West Spitsbergen Current waters. The lateral stations at the trough, i.e., those in proximity to the fjords sites, showed cold Surface Water (southern Sta. III: T=−0.2C,S = 34.6; northern Stas.
VI and VII: T = 0.7C, S = 34.8), which turned into slightly warmer (2.2C) and more saline water (34.9) at 35 m water depth, indicating that Surface Water lost more heat to the atmosphere than deeper waters, and that water was still mixing with the West Spitsbergen Current waters; Trans- formed Atlantic Water was dominating the deep-water col- umn down to 365 m water depth (> 34.86, 1.6–2.7C). Further inside Isfjorden (Stas. I and II), close to Longyearbyen, the entire water column was comprised of Local Water with salin- ity of 34.8 and temperatures of 0.8–1.2C (Fig. S1A).
A B C D
E F G H
I J K
L M N O
May (2017)June (2016)May (2016)July (2015)
Eastern N-S Western N-S Southern W-E Northern W-E
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
450
400
350
300
250
200
150
100
50
0
Methane [nM]
Water depth [mbsl]
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 0 5 10 15 Distance along transect [km]
Fig 3.Profiles of dissolved CH4in the water column along transects at the shallow shelf west of Prins Karls Forland (A–O) from four sampling surveys in May, June, and July within three successive years (2015–2017). For each sampling survey, water depth on they-axis is given in meters below sea level (mbsl). White circles represent single water samples. The color code shows measured and linearly interpolated CH4concentrations (nM). Each plot con- tains the bathymetrical baseline (black line above gray area) characteristic for each transect. In June 2016, only three transects were conducted.
In June 2016, Isfjorden surface waters down to 40 m were considerably fresher and warmer (34.2, 4.7C) than in May 2016, most likely due to freshening through melting of gla- ciers, snow and sea ice and heating from the atmosphere (Fig. S1B).
Compared to Isfjorden in June 2016, a similar structuring of the water column was observed outside Bellsundet. In late June, surface temperatures were ~4C, heated up by the atmo- sphere and surface salinities of 34.3–34.65 indicate the influ- ence of Arctic Water, carried by the East Spitsbergen Current.
Similarly, at Outer Hornsund, southwards from Bellsundet, surface salinity was not higher than 34.6, indicating the pres- ence of Arctic Water transported by the East Spitsbergen Cur- rent (Fig. S1B).
Water column methane content
The entire water column was CH4 oversaturated with respect to the atmospheric equilibrium concentration, which is ~ 3 nM at the ambient salinity and temperature conditions (Wiesenburg and Guinasso (1979) (Table S2). In the area west of Prins Karls Forland, we observed CH4plumes with concen- trations of up to 437 nM. High concentrations were mainly encountered in bottom waters within the flare cluster, most dominantly in the southwest of the sampling grid, but the extent of the plumes differed greatly between the surveys (Figs. 3, S2). For example, we found elevated CH4 concentra- tions extending widely from west to east in May 2017 (Fig. 3E–H) and June 2016 (Fig. 3I–K). In contrast, the eastward extension was less pronounced in May 2016 (Fig. 3A–D) and a clearly defined CH4plume was located at the intersection of the southern W-E and the western N-S transect (Fig. 3B,C).
Nevertheless, the mean content of dissolved CH4in the water column was similar when comparing the different surveys (3483, 3547, and 3745μmol m−2in May 2016, May 2017, and July 2015, respectively) (Table 2, Fig. S2). In June 2016, the mean content of dissolved CH4 reached 5644 μmol m−2 due to the reduced number of stations (12 out of 64 stations) that were sampled during this survey, and that many of the sam- pled stations were located above active flares (see sampling strategy), which in turn translated to higher mean values (see the calculated mean content of dissolved CH4for the reduced number of stations for all surveys in Table S2). Therefore, dis- solved CH4values from June 2016 are not directly comparable to values from the other surveys.
In general, CH4 concentrations were highest in bottom waters, translating to inventories that were also highest at the Bottom Water Layer (2127–2867μmol m−2) compared to the Intermediate (795–1008 μmol m−2) and Surface Water Layer (83–412μmol m−2) (Table 2).
At Isfjorden (Stas. I and II) we observed elevated concentra- tions of 26 and 57 nM (May and June 2016, respectively) in the Bottom Water Layer, whereas at the Isfjorden Trough CH4
was generally low, with average values of 9 nM in the Bottom and 3 nM in the Surface Water Layer (May 2016). At Outer Bellsundet, Outer Hornsund and Sørkappøya, CH4concentra- tions in the Bottom Water Layers were 18, 12, and 24 nM, respectively, and 11, 4, and 17 nM in surface waters (Table S2).
Methane oxidation activity
Highest MOx activity was generally found in bottom waters, although the magnitude of activity greatly varied
Table 2.Inventory of dissolved CH4and microbial methane oxidation activity calculated for the sampling area at the shallow shelf of Prins Karls Forland. Surface, Intermediate, and Bottom refers to the defined water layers (Table 1) of the water column. Total values are the sum of all three water layer values per sampling campaign. The order of sampling campaigns in this table follows the cycle of the seasons where May corresponds to Arctic spring and July to summer.
Dissolved methane Methane oxidation activity
CH4oxid. per day* 100% turnover**
Surface Interm. Bottom Total Surface Interm. Bottom Total
Mean content (μmol m−2) Mean turnover (μmol m−2d−1) (%) (d)
May (2016) 100 928 2456 3483 0.006 0.18 0.33 0.51 0.015 6777
May (2017) 412 1008 2127 3547 0.104 0.48 1.46 2.02 0.057 1754
July (2015) 83 795 2867 3745 0.089 1.73 25.72 27.54 0.735 136
Dissolved methane Methane oxidation activity
Surface Interm. Bottom Total Surface Interm. Bottom Total
Total content in the area (×105mol) Total turnover in the area (mol d−1)
May (2016) 0.37 3.38 8.94 12.68 2 65 120 187
May (2017) 1.50 3.67 7.74 12.91 38 173 532 736
July (2015) 0.30 2.89 10.44 13.63 32 630 9362 10,024
*Percentage of CH4that is oxidized per day.
**Time in days that it would take to totally oxidize the available CH4.
between surveys (Figs. 4, S3). We found a MOx maximum value of 7.1 nM d−1in the south-east part of the sampling area at the Prins Karls Forland shelf in July 2015 (Sta. 53; Fig. S4).
In contrast, MOx maxima in May 2016, May 2017 and June 2016, were more than an order of magnitude lower with values of 0.09, 0.16, and 0.23 nM d−1, respectively (Fig. S4).
Similarly to maximum rates, depth integrated MOx activity in the Bottom Water Layer was also highest in July 2015 (25.72 nmol m−2 d−1) and substantially lower in May 2016 (0.33 nmol m−2d−1), May 2017 (1.46 nmol m−2d−1) and June 2016 (1.67 nmol m−2d−1). In the Surface Water Layer, average MOx activity was generally below 0.1 nmol m−2d−1(Table 2, Fig. S3).
Among the stations along the transect inside Isfjorden, highest MOx activity was in the bottom waters at Sta. I
(0.04 nM d−1in May and 0.5 nM d−1in June 2016 (Fig. S4);
no samples were taken in Isfjorden in 2015). MOx rates at Outer Bellsundet, Outer Hornsund and Sørkappøya wer- e < 0.04 nM d−1(Table S2).
Methanotrophic community
The particulate methane monooxygenase gene (pmoA) was sequenced from selected bottom water samples with elevated MOx rates. The selected samples originated from theflare area (Sta. 9 May 2016 and 10 July 2015), the bathymetric depression zone (Stas. 54 May, June 2016, July 2015 and 58 May 2016, July 2015), Sta. 49 located at the north-east corner of the sampling grid (May 2016 and July 2015), Sta. I at Isfjorden (May and June 2016), and from Outer Bellsundet (Sta. XI June 2016) (Fig. 1, Table S1). The number of generated pmoA sequences ranged
May (2017)June (2016)May (2016)July (2015)
Eastern N-S Western N-S Southern W-E Northern W-E
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0
Water depth [mbsl]
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 0 5 10 15 Distance along transect [km]
A B C D
E F G H
I J K
L M N O
7 6 5 4 3 2 1 0
MOx [nM d-1]MOx [nM d-1]
Fig 4.Profiles of microbial CH4oxidation (MOx) rates in the water column along transects at the shallow shelf west of Prins Karls Forland (A–O) from four sampling surveys in May, June, and July within three successive years (2015–2017). For each sampling survey, water depth on they-axis is given in meters below sea level (mbsl). White circles represent single water samples. The color code shows measured and linearly interpolated MOx rates (nM d−1). Each plot contains the bathymetrical baseline (black line above gray area) characteristic for each transect. In June 2016, only three transects were conducted. Note that we have used the same interpolation settings for all transects (with a resolution of 0.1 nM d−1), but have chosen two color scales for the July vs. May and June expeditions.
from 20,132 to 44,469 per sample. After processing, quality reads clustered into 70 OPUs with absolute abundance of maximum 34,299 and minimum 424 reads across samples. The most abun- dant OPUs were related to the gammaproteobacterial deep-sea 3 and deep-sea 1 clades, both of which are subgroups of gammaproteobacterial Methylococcaceae Type Ia MOB according to Lüke and Frenzel (2011) (Figs. 5, S6). In bottom water samples from May 2016, the relative abundance of Type Ia deep-sea 1 MOB was higher compared to the other months. In addition, OPUs belonging to unclassified Proteobacteria or Methylococcaceae-related genera only showed low sequence abundance.
Bacterial diversity
Our 16S rRNA gene analyses of water column bacterial community compositions revealed a great phylogenetic diver- sity and spatial variability. After sequence processing of > 1.2 million raw sequences from 57 samples, 11,705 OTUs were generated.
The majority of the 16S rRNA gene sequences clustered into OTUs which are taxonomically affiliated with Alphaproteobacteria (34%), Gammaproteobacteria (30%), Bacteroidetes (25%), and Verrucomicrobia (4%) (Fig. 6B). Among the Alphaproteobacteria, the most abundant families were SAR11 clade I, SAR11 clade II
and Rhodobacteraceae (PlanktomarinaandSulfitobacter). Rela- tives of Gammaproteobacteria were mainly affiliated with Nitrincolaceae, Thioglobaceae, SAR86 clade, Porticoccaceae, and Methylophagaceae. The majority of the Bacteroidetes sequences were classified as Flavobacteriaceae with the domi- nant genera Polaribacter 1, Polaribacter, NS5 marine group, and Aurantivirga. Other abundant Bacteroidetes were NS9 marine group, Cryomorphaceae and Bacteroidaceae.
Among the Verrucomicrobia, Rubritaleaceae was the most abundant family. Sequences affiliated with Luteolibacter and Roseibacilluswere present in low amounts in almost all sam- ples, but slightly more abundant in bottom waters sampled in July 2015.
Only few 16S rRNA gene sequences were associated with known methanotrophic or methylotrophic bacteria OTUs (related to either Alpha- or Gammaproteobacteria; Fig. 6D).
Known MOB found in the data set were related to clade Milano-WF1B-03 (6 OTUs; Heijs et al. (2005)), found in samples from bottom waters at Stas. 54 and 58 in July 2015 and at station IF in May and June, and to the Meth- yloprofundus clade (4 OTUs) found at Stas. 9 and 19 sampled in May 2016. Furthermore, OTUs affiliated to Methylobacterium, Methyloceanibacter, and unclassified Methylomonaceae (one OTU each) were identified.
10 20 30 40 50 60 70 80 90 100 0.05
0.10
0.15 0
A B
58_1 May (2016) 54_1 May (2016) 9_1 May (2016)
I_1 May (2016) 49_1 May (2016) I_1 June (2016) 54_1 June (2016) XI_2 June (2016) 49_1 July (2015) 54_1 July (2015)
10_1 July (2015) 58_1 July (2015)
AW / TAW AW
TAW TAW
Type Ia deep-sea 3 (OTU1) Type Ia deep-sea 1 (OTU2) Type Ia PS-80
MOB-like Bacteria Methylomonas-related Methylococcaceae
Methylococcales Gammaproteobacteria Proteobacteria
Others < 1%
Methylococcales
Hierarchical cluster dendrogram [weighted UNIFRAC distance] Relative abundunce of pmoA gene sequences [%]
Unclassified
Fig 5. Hierarchical clustering of OPUs derived frompmoA gene sequences (A) and relative abundance (B) of the methanotrophic community from selected stations from the shallow shelf of Prins Karls Forland, Isfjorden (I) and Outer Belsundet (XI) investigated over three sampling surveys in May 2016, June 2016, and July 2015. Sample IDs derive from the station number (see sampling grid in Fig.1B) and water levels (1: 5 m above seafloor, 2:
15 m above seafloor). Gray squares show predominant water masses found at the bottom water level at the stations (AW: Atlantic Water, TAW: Trans- formed Atlantic Water).
Methylotrophs were represented by members related to uncultured Methylophagaceae (110 OTUs in total) present in samples from the shelf west off Prins Karls Forland taken in June 2016 and July 2015; one OTU was identified as Methylophagaand found at Sta. I, Isfjorden.Methylotenera(two OTUs) was encountered in the whole water column in July 2015 and other genera of the family Methylophilaceae (OM43 clade, 72 OTUs) were present at all seasons. Known MOB accounted for 0.05% of all sequences, and known meth- ylotrophs accounted for 1.08% of all sequences (Fig. 6D).
Community beta diversity analysis revealed the time of sam- pling (sampling campaign) as one dominant factor shaping the bacterial community composition (envfit,p≤0.05). The dissimi- larity of the bacterial community was particular apparent in sam- ples retrieved in May 2016 (spring) compared to samples from June 2016 and July 2015 (late spring/summer; Fig. 7A). In addi- tion to the sampling campaign, water depth was a second vari- able that significantly correlated with community dissimilarity (p≤0.05). To reduce the masking of seasonal effects on the beta diversity analysis, we subsequently focused on samples retrieved from single sampling campaigns. Here, water temperature and depth, both independent environmental variables, influenced the communities (p ≤0.05). Since these factors/variables (together with salinity) also define the classification of water masses (see section Hydrographic setting), water masses are indi- cated in Fig. 7B–D. In June 2016 and July 2015, communities revealed similarities according to water masses (Fig. 7C,D). Anal- ogously, medium or high active MOB communities were more similar to one another. At Isfjorden and Outer Bellsundet, where water mass properties were highly affected by local features, communities were distinctively different to most of the other communities found at the shelf west off Prins Karls Forland, especially in June 2016 (Fig. 7C).
Supplementing NMDS-based analysis, we conducted canoni- cal correspondence analysis (CCA). Similar to NMDS, CCA also indicated that water temperature, depth, and salinity signifi- cantly influenced the community composition (Table 3). For the bacterial communities in May and June 2016, these environ- mental variables adequately described the variation of the com- munity composition, as supported by a significant level for the CCA Model (ANOVA;p≤0.009 andp≤0.002, respectively). In contrast, the composition of samples retrieved in July 2015 showed a much higher variation than could adequately be explained by the investigated environmental variables included in the model (p≤0.129), suggesting that additional unidentified factors played a major role. When taking community-dependent variables into account, such as amount of extracted DNA and CH4oxidation, both correlated with the identified communities (Table 3).
To identify a possible correlation of bacterial phyla with methane oxidation rates, we conducted a Spearman’s rank cor- relation analysis on family level. Following clades depicted the greatest positive correlations: unclassified members of the OCS116 clade, Nitrosomonadaceae, Cellvibrionaceae, clade
OM182, clade ZD0405, Thiothrichaceae, Rubritaleaceae, and Verrucomicrobiaceae (Fig. S7). Many of these families also depicted a positive correlation with CH4 concentration and water level. Methane concentration, water depth and methane oxidation rates were also strongly correlated with another.
Repeated sampling
Hydrographical parameters (salinity, temperature, pressure), concentration of dissolved CH4and MOx activity were repeat- edly measured at Stas. 9, 16, 31, 44, 54, and 64 over a 2-day time period (Table S1). Water mass properties only showed marginal differences (Fig. S8A,D). Stations located above or close to theflare area (Stas. 9, 19, 31, and 44) showed stronger variations in CH4 concentrations in samples from greater water depths. MOx activity rates from all six stations (Fig. S8C,F) in addition to the 16S rRNA gene sequencing results from Sta. 9, showed high similarities when comparing the two time points (Fig. 6B).
Discussion
The shallow shelf west off Prins Karls Forland is character- ized by numerous gas flares at the ridge of the Forlandet moraine complex as well as the many bathymetric depressions extending eastwards from the moraine. The water column at the shallow shelf is a hydrographically complex and dynamic system with seasonal variations in water mass properties. Indi- vidual gas flares transport differential amounts of CH4 into the water column, and total CH4flux on the shelf also varies over time (Silyakova et al. 2020). However, a seasonal connec- tion with high CH4fluxes during the warm season and ~ 80%
lowerfluxes during cold bottom water conditions, as found at the shelf break below 360 m water depth (Ferré et al. 2020), is not evident on the shallow shelf, where active flare clusters occur at 90 m. Such a depth is far above the uppermost limit of the shifting gas hydrate stability zone, which was found to be in between 380 and 400 m water depth (Berndt et al. 2014). We repeatedly investigated the shallow shelf over a time period of 3 years covering the Arctic spring (May 2016, 2017), late spring (June 2016) and summer (July 2015) and their specific hydrographic conditions. Our study reveals the activity, distribution and structure of methane-oxidizing com- munities in the water column on the shallow shelf west of Svalbard.
Spatiotemporal variations of methane content in the entire water body
Similar to previous studies on CH4 dynamics in coastal waters of Svalbard (e.g., Damm et al. 2005; Graves et al. 2015;
Mau et al. 2017), we generally observed highest CH4concen- trations in bottom waters. In our sampling grid west of Prins Karls Forland (Fig. 1B), CH4 concentrations frequently exceeded 100 nM in particular at gasflares locations (Fig. 3).
Methane concentrations in surface waters were supersaturated compared to atmospheric concentrations across all surveys
0 10 20 30 40 50 60 70 80 90
100 49_8 I_3 54_5 19_3 19_8 54_8 9_8 I_1 58_5 58_8 19_1 9_5 9*
_1 49_1 58_1 54_1 54_3 9_3 9*_8 9_1 19_5 49_3 49_5 9*_3 9*_5 XI_4 XI_2 I_1 I_3 49_8 10_1 49_1 10_3 54_8 49_5 54_5 10_8 49_3 54_3 54_1 10_5 49_8 49_5 49_1 54_3 58_3 54_1 58_1 49_3 10_1 10_3 58_5 10_5 54_5 58_8 10_8 54_8
0 0.2 0.4 0.6
Relative abundance of 16S rRNA gene sequences [%] 49_8 I_3 54_5 19_3 19_8 54_8 9_8 I_1 58_5 58_8 19_1 9_5 9*_1 49_1 58_1 54_1 54_3 9_3 9*_8 9_1 19_5 49_3 49_5 9*_3 9*_5 XI_4 XI_2 I_1 I_3 49_8 10_1 49_1 10_3 54_8 49_5 54_5 10_8 49_3 54_3 54_1 10_5 49_8 49_5 49_1 54_3 58_3 54_1 58_1 49_3 10_1 10_3 58_5 10_5 54_5 58_8 10_8 54_8
0.4
0.3
0.2
0.1
0
Portion of 16S rRNA sequences [%]
A
B
DC
*
*
*
* *
* *
* * * *
*
* *
* * * * * *
May (2016) June (2016) July (2015)
Other Verrucomicrobia Rubritaleaceae
SAR324 Marine Group B Other γ-Proteobacteria
Thiotrichaceae Thioglobaceae SAR86 clade Moraxellaceae OM182 clade
Other α-Proteobacteria SAR11 unclassified SAR11 clade II SAR11 clade I Rhodobacteraceae
Marinimicrobia SAR406 Carnobacteriaceae Other Firmicutes Other Bacteroidetes NS9 marine group Flavobacteriaceae Cryomorphaceae Bacteroidaceae Other Actinobacteria Microtrichaceae Phylum Verrucomicrobia
Planctomycetaceae Phylum δ-Proteobacteria Class α-Proteobacteria Class γ-Proteobacteria
Phylum Actinobacteria Phylum Bacteroidetes Phylum Firmicutes Phylum Marinimicrobia
Others Class γ-Proteobacteria
Methyloceanibacter Methylobacterium Milano-WF1B-03 Methyloprofundus Methylomonaceae uncl.
Class α-Proteobacteria Legend D
Legend B
*
Stations above gas flaresMOx activity (nM d-1) high >0.5 medium 0.5-0.1 low 0.05-0.005 no <0.005 Legend C Pseudohongiellaceae
Nitrincolaceae Methylophagaceae Porticoccaceae Methylophilaceae Colwelliaceae Phylum Planctomycetes
Fig 6.Legend on next page.
(1–5 fold on average). However, substantial impacts on atmo- spheric CH4concentrations in the region of the western Sval- bard continental margin could not be confirmed (Myhre et al. 2016).
Furthermore, the integrated CH4 inventory of the three water layers of the sampling grid during the surveys in spring (May 2016 and 2017), late spring (June 2016) and summer (July 2015), shows up to three times higher CH4values at the Bottom Water Layer compared to the Intermediate and Sur- face Water Layers, as a result of the CH4seepage from the sea- floor (Table 2, Fig. S2). Moreover, the total content of dissolved CH4in the area (423 km2) is consistent with values of ~ 13×105mol area−1in spring (May 2016/17) and summer (July 2015). The seemingly high total CH4 content in late spring (June 2016) is caused by an under-sampling of the sur- vey area (see section Water column methane content and Study area and Sampling strategy).
Our values from May 2016/17 and July 2015, together with previously publishedflux data (Silyakova et al. 2020), indicate comparably steady CH4 inputs in our study area during the investigated seasons. In contrast, seep activity in deeper water levels at the shelf break are strongly reduced during times with low bottom water temperatures (Ferré et al. 2020), because the uprising CH4”freezes out”as gas hydrate in surface sediments, building up a seasonal gas hydrate capacitor that is reduced in summer.
Moreover, our repeated sampling over the short time period of 2 days in May 2016 showed negligible variations in CH4con- centration and water mass properties (Fig. S8A,D), and neither the bacterial MOx activity (Fig. S8C,F) nor the community com- position (Fig. 6B) revealed any remarkable differences between the two time points. Thesefindings indicate that hydrographic and biogeochemical variations during one sampling of the entire grid were most probably low (Steinle et al. 2015).
Spatiotemporal variations of methane oxidation activity Methane oxidation in the ocean is the final sink for dis- solved CH4 before its release into the atmosphere (e.g., Reeburgh 2007; Steinle et al. 2015). Previous studies report that elevated MOx activity in marine environments is related to high CH4 concentrations (Valentine et al. 2001;
Mau et al. 2013; Crespo-Medina et al. 2014; Steinle et al. 2015). In our study, we also found elevated MOx rates in methane-rich bottom waters. But in the CH4 plumes, MOx was not substantially elevated (Figs. S2, S3). This has been
found elsewhere, too (Crespo-Medina et al. 2014; Steinle et al. 2015, 2017), and a literature review only revealed a cor- relation of MOx and CH4 contents on logarithmic scales (James et al. 2016). The rather loose dependency of MOx and CH4 concentrations indicates that microbial community abundance, and possibly other factors such as the availability of micronutrients, seems at least equally important in deter- mining the efficacy of the microbial CH4 filter in the water column (Steinle et al. 2015).
Our study area is characterized by steady CH4 contents between seasons, but similarly to the spatial variation of MOx within one sampling campaign, we found large seasonal dif- ferences in MOx activity. In the Arctic spring (May) and late spring (June), MOx rates were generally low (weighted mean:
< 2.02 μmol m−2d−1; total MOx: < 736 mol d−1; Table 2). In contrast, in summer (July), MOx in the entire area was about one order of magnitude higher (weighted mean:
27.54 μmol m−2 d−1; total MOx: 10,024 mol d−1). It is also noteworthy that the maximum MOx value measured in sum- mer (July; 7.2 nM d−1) was much higher compared to previous measurements conducted in the area around Svalbard. Steinle et al. (2015) measured MOx rates of up to 3.2 nM d−1at the continental slope west of our study area in Arctic summer (August). Mau et al. (2017) published rates of up to 2.2 nM d−1 in a CH4 plume located more southerly between Hornsundbanken and Isfjordbanken west of Spitsbergen from the same season (August/September).
We discovered that the capacity of MOx shows a high spa- tiotemporal variation. The high MOx rate in summer trans- lates to a turnover time of the CH4 inventory of the entire sampling grid (13.63×105 mol) of about 136 d. In contrast, the turnover time was substantially longer in spring (1754–6777 d). While MOx plays a substantial role in retaining CH4in the Arctic summer, it seems of rather lower importance in winter. Similar seasonal differences were also found at the shelf break west of our study area (Steinle et al. 2015; Ferré et al. 2020). In general, the turnover times at the Prins Karls Forland shelf are within the intermediately high to low range when compared to previously reported turnover times of weeks to a few years from methane-rich, Arctic waters (Mau et al. 2013; Steinle et al. 2015; James et al. 2016). Turnover times of several decades are rare and typically restricted to oceanic deep waters with very low CH4
contents (< 10 nM) (Rehder et al. 1999; Heeschen et al. 2003;
James et al. 2016).
FIGURE 6 Differential bacterial community structure based on 16S rRNA genes investigated over three sampling surveys of the shelf west of Prins Karls Forland, Isfjorden and Outer Belsundet. (A) Hierarchical clustering of bacterial communities of each sampling survey is based on subsampled Bray–Curtis dissimilarity matrix (OTU) and the complete linkage method. Stability of clusters was tested by bootstrapping 1000 times. (B) Relative abundance of bac- teria based on 16S rRNA gene sequences are sorted according the hierarchical clustering within each sampling survey. Only taxa with abundances of
>1% of total sequences are shown. (C) Simplified ranking of measured methane oxidation (MOx) activities per sample. (D) Proportion of 16S rRNA sequences, which were assigned to methanotrophic bacteria. Sample IDs are derived from the station number (see sampling grid in Fig. 1B) and water levels (1: 5 m above seafloor, 3: 25 m above seafloor, 5: Intermediate water level, 8: 5 m below sea surface, I: Isfjorden, XI: Outer Belsundet). Sta. 9 was repeatedly sampled during the sampling campaign; repeated samples are therefore marked with asterisks (9*_1 to 8).
Hydrographical dynamics on the shelf
The spatiotemporal variations of CH4 content and MOx activity indicate two contrasting mixing regimes at the shelf, both of which have profound effects on MOx activity as well as the bacterial community composition.
Thefirst scenario is characterized by a water column domi- nated by Atlantic Water as it was typically the case in summer (July; Fig. 2L–O). Atlantic Water episodically floods the shal- low shelf in the form of numerous eddies caused by the West Spitsbergen Current that meanders eastwards onto the shelf (Nilsen et al. 2008; Steinle et al. 2015). The dense Atlantic Water replaces the shelf water, less saline (though colder) Arc- tic Water brought by the East Spitsbergen Current, andfills up the bathymetric depressions (Silyakova et al. 2020). This phe- nomenon was particularly apparent at the eastern end of the southern W-E transect, where the depressions are 40 m deeper
than the surrounding seafloor. There, we found hot spots of MOx activity with 2–3 times higher rates than those reported previously from the continental shelf around Svalbard (Mau et al. 2013; Gentz et al. 2014; Steinle et al. 2015, 2017) although CH4 concentrations were only moderately high in the depressions compared to gasflare locations in the western part of the sampling grid (Figs. S2, S3). Prior to flooding the shelf, Atlantic Water has an offshore history where CH4con- centrations are low (Steinle et al. 2015). When swept over the CH4 seeps at the shelf break (i.e., west of the study area), Atlantic Water becomes charged with CH4, but MOx rates in the water column are initially low because of the initially low MOB content in this water mass (Steinle et al. 2015). When reaching the depressions, methane-enriched Atlantic Water becomes trapped as these depressions provide a sheltered environment with long residence times. This supports MOB
NMDS2NMDS2
NMDS1 NMDS1
July (2015) June (2016) May (2016)
high rMOx medium rMOx low rMOx no rMOx
TAW LW
May (2016)
June (2016) July (2015)
high rMOx medium rMOx low rMOx no rMOx
high rMOx medium rMOx low rMOx no rMOx
A B
C D
I-3
I-1
I-3 I-1 XI-4
XI-2
Stress: 0.09
Stress: 0.12
Stress: 0.05 Stress: 0.14
TAW LW AW
Fig 7.Seasonal correlation of microbial communities from all samples retrieved from all sampling surveys (A) and between microbial communities and salinity (red lines and values) according to single sampling campaigns (B–D). Non-metric multidimensional scaling (NMDS) derive from the Bray–Curtis dissimilarity index. AW: Atlantic Water; TAW: Transformed Atlantic Water; LW: Local Water; IW: Intermediate Water; I: Isfjorden (water levels 1 and 3); XI:
Outer Bellsundet (water levels 2 and 4). Methane oxidation rates (rMOx) are defined as high: >0.5, medium: 0.5–0.1, low: 0.05–0.005, no:
<0.005 nM d−1.
Table 3.Canonical correspondence analysis (CCA) significance values of independent and dependent variables. Values marked with * indicate values of significance. Temp: temperature; Fluor:fluorescence; CH4: dissolved methane concentrations; DNA: 16S rRNA gene sequencing analysis; MOx: methane oxidation rates. The order of sampling campaigns in this table follows the cycle of the Arctic sea- sons where May corresponds to spring, June to late spring, and July to summer.
Independent variables Dependent variables
ANOVA, CCA model Temp. Depth Salinity Fluor. CH4 DNA MOx
May (2016) 0.009* 0.005* 0.050 0.315 0.155 0.275 0.560 0.030*
June (2016) 0.002* 0.005* 0.005* 0.150 0.015* 0.580 0.025* 0.005*
July (2015) 0.129 0.025* 0.005* 0.005* 0.465 0.165 0.005* 0.035*
